When research is conducted across several languages, the quality of the findings depends partly on how well the questions and responses travel between markets. A translation can be grammatically correct and still change the way a respondent understands a question, an answer scale or a product concept.

These problems are not always obvious. The survey may run without technical issues and the final dataset may look complete. However, if respondents in different markets have interpreted the material differently, the results may no longer be directly comparable.

This is why market research translation should be treated as part of the research methodology, rather than as a final administrative step.

How can translation affect market research data?

Research questions are carefully designed to measure attitudes, preferences and behaviour. Small changes in wording can alter that measurement. A term such as “good value”, “easy to use” or “likely to recommend” may carry a different level of intensity or formality in another language.

The risk is not limited to the question itself. Answer options, routing text, piped responses, instructions and stimulus materials all contribute to the respondent experience. If one part is unclear or inconsistent, it can influence how people answer.

In qualitative research, meaning can be affected in a different way. Interviews and focus groups often contain humour, hesitation, emotion, idioms and cultural references. A literal translation may capture the words but lose the point the participant was making.

Common points where meaning can be lost

1. Translating a questionnaire without enough context

A linguist needs to know who the respondents are, what the study is trying to measure and how the questionnaire will be used. Without that context, an ambiguous phrase may be translated in a way that is linguistically reasonable but unsuitable for the research objective.

2. Changing the meaning of response scales

Response scales depend on clear distinctions. Terms such as “somewhat”, “very” and “extremely” need to retain an appropriate progression in every language. If two options become too similar, or one sounds much stronger than intended, the distribution of responses can change.

3. Using inconsistent terminology

Product names, category terms and recurring concepts may appear across screeners, questionnaires, discussion guides and stimulus materials. Inconsistent translation can confuse respondents and make later analysis more difficult. A project glossary and agreed terminology help reduce this risk.

4. Checking the file but not the live survey

A translated spreadsheet can be accurate while the programmed survey still contains problems. Text may be cut off, routing may display the wrong wording, links may fail or piped text may become ungrammatical. Overlay and link checking allows the live respondent experience to be reviewed before launch.

5. Oversimplifying open-ended responses

Verbatim responses are rarely neat. They may include spelling errors, shorthand, regional language or incomplete thoughts. Translating them too literally can remove tone or imply a certainty that was not present in the original. This can affect subsequent coding and content analysis.

A practical quality process for multilingual research

There is no single workflow that suits every study. The right level of review depends on the audience, subject matter, number of markets, research method and consequences of an error. A carefully managed process will usually include the following steps.

1. Provide a clear brief. Explain the research objective, respondent profile, intended tone, delivery format and any sensitive or regulated terminology.
2. Use linguists who understand market research. Familiarity with questionnaires, discussion guides and respondent language helps linguists recognise issues that a general translation approach may miss.
3. Agree terminology early. A glossary or reference list supports consistency across languages, files and project stages.
4. Use an appropriate review method. This may involve revision by a second linguist, team review, adjudication or back translation, depending on the study. The Cross-Cultural Survey Guidelines explain why review should assess meaning and measurement, not only word-for-word similarity.
5. Check the live environment. Review survey layout, logic, links and local-language display after programming.
6. Document decisions. Recording linguistic queries and approved wording helps later waves of a tracker study remain consistent.

The MRS and ATC checklist for buying translation services also highlights the value of supplier selection, quality-managed processes and checks that are appropriate to the research project.

Where can AI and machine translation fit?

Technology may support some market research translation workflows, particularly when teams need rapid internal understanding of a large volume of low-risk content. It should not be assumed to be suitable for every questionnaire, transcript or open-ended response.

Respondent-facing and insight-critical material often contains ambiguity, cultural context and specialist terminology that require human judgement. Any use of machine translation should therefore be agreed with the client, assessed for the project and supported by suitable quality control and secure handling.

Our guide to AI in market research translation looks at these decisions in more detail.

Better language decisions support better research decisions

Poor translation does not always produce an obvious failure. More often, it introduces small differences that are difficult to see once the data has been collected. Careful briefing, specialist linguists, terminology control and live checks help research teams identify those risks before they affect respondents.

Language Insight supports multilingual research through survey translation, translation, transcription, overlay and link checking, coding, content analysis and simultaneous translation. We work closely with clients to choose a process that is appropriate for the study, its audience and the intended use of the findings.

If you are planning a multilingual study, contact our team to discuss the language stages and quality checks that would best support it.

Frequently asked questions

Can translation really change survey results?

Yes. Changes in meaning, tone or response-scale intensity can affect how respondents interpret and answer a question. The effect may be subtle, which is why contextual review and testing are important.

What is market research translation?

Market research translation covers materials used to recruit participants, collect responses and analyse findings across languages. These may include screeners, questionnaires, discussion guides, stimulus materials, transcripts, verbatim responses and reports.

Is back translation always necessary?

No. Back translation can be useful for sensitive or high-risk studies, but it is not automatically the best method for every project. The review approach should reflect the study design, subject matter and level of risk.

Why should a live multilingual survey be checked?

Programming can introduce issues that are not visible in the translation file, including truncated text, broken links, incorrect routing and awkward piped wording. A live check reviews what respondents will actually see.

Can machine translation be used for open-ended responses?

It may be suitable for some low-risk or internal-use datasets, but open-ended responses often contain informal language, ambiguity and cultural references. Suitability, security and the need for human review should be assessed before the workflow is agreed.